Dew Deposition Suppresses Transpiration and Carbon Uptake in Leaves T ⁎ Cynthia Gerlein-Safdia, , Michael C

Agricultural and Forest Meteorology 259 (2018) 305–316

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology

journal homepage: www.elsevier.com/locate/agrformet

Dew deposition suppresses transpiration and carbon uptake in leaves T ⁎ Cynthia Gerlein-Safdia, , Michael C. Koohafkanb,c, Michaella Chungc, Fulton E. Rockwelld, Sally Thompsonc, Kelly K. Caylora,e,f a Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08544, USA b Hydrologic Sciences Graduate Group, University of California Davis, Davis, CA 95616, USA c Department of Civil and Environmental Engineering, University of California Berkeley, Berkeley, CA 94720, USA d Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138, USA e Department of Geography, University of California Santa Barbara, Santa Barbara, CA 93106, USA f Bren School of Environmental Science and Management, UC Santa Barbara, Santa Barbara, CA 93106, USA

ARTICLE INFO ABSTRACT

Keywords: Dew deposition occurs in ecosystems worldwide, even in the driest deserts and in times of drought. Although Carbon assimilation some species absorb dew water directly via foliar uptake, a ubiquitous effect of dew on plant water balance is the Dew interference of dew droplets with the leaf energy balance, which increases leaf albedo and emissivity and de- Fog creases leaf temperature through dew evaporation. Dew deposition frequency and amount are expected to be Leaf energy balance affected by changing environmental conditions, with unknown consequences for plant water stress and eco- Leaf wetness system carbon, water and energy fluxes. Here we present a simple leaf energy balance that characterizes the Transpiration suppression effect of deposition and the evaporation of dew on leaf energy balance, transpiration, and carbon uptake. The model is driven by five common meteorological variables and shows very good agreement with leaf wetness sensor data from the Blue Oak Ranch Reserve in California. We explore the tradeoffs between energy, water, and carbon balances for leaves of different sizes across a range of relative humidity, wind speed, and air temperature conditions. Our results show significant water savings from transpiration suppression up to 25% for leaf char-

acteristic lengths of 50 cm. CO2 assimilation is decreased by up to 12% by the presence of dew, except for bigger − leaves in windspeed conditions below 1 m s 1 when an increase in assimilation is expected.

1. Introduction water-repellent leaves (Neinhuis and Barthlott, 1997) that are adapted to shed water rather than trapping and imbibing it. Non-meteoric water Non-meteoric water (dew or fog) is an often-overlooked component deposition can impose risk on plants: in cold climates it may freeze, of surface water balance for many landscapes. Dew in particular can damaging leaf tissue (Jordan and Smith, 1994), and in warm environ- occur in virtually every climate and ecosystem worldwide, from the ments it may cause rotting and facilitate pathogen infection (Evans arid Negev Desert of Israel (Zangvil, 1996; Hill et al., 2015), to flooded et al., 1992). As a response, many species have developed waxy, hy- rice paddies of China (Xu et al., 2015), semi-arid costal steppes of Spain drophobic cuticles (Holloway, 1994) that force water droplets to bead (Uclés et al., 2013), American wheat fields (Pinter, 1986), lush up and roll off the leaf. However, the capacity for droplets to drain off Northern European lawns (Monteith, 1957; Jacobs et al., 2006), and the leaf can be very small (Gao and McCarthy, 2006), in spite of high tropical forests (Clus et al., 2008; Lakatos et al., 2012). Non-meteoric contact angles between droplet and leaf surface (Holder, 2012b). water can affect the water balance of plants directly through foliar Therefore, during dew events, micro-droplets of water may sometimes uptake (Yates and Hutley, 1995; Andrade, 2003; Lakatos et al., 2012; form on the surface of even hydrophobic leaves (Holder, 2012b; Berkelhammer et al., 2013), and can act as a major water source for Aparecido et al., 2017). vegetation where fresh water is scarce (Agam and Berliner, 2006; Clus The accumulation of dew droplets on leaf surfaces directly affects et al., 2008). Some species that lack access to soil water – including leaf energy balance through evaporative cooling, changes in albedo, epiphytic bromeliads (Andrade, 2003; Gotsch et al., 2015; Pan and and changes in emissivity. These factors – and their interactions – lead Wang, 2014) and lichens (Lakatos et al., 2012) – have physical features to transpiration suppression – a collective term that describes reduc- allowing them to collect dew water. Most plant species, however, have tions in leaf-level transpiration (Tolk et al., 1995; Gerlein-Safdi et al.,

⁎ Corresponding author at: University of Michigan, Department of Climate and Space Sciences and Engineering, 2517C SRB, 2455 Hayward Street, Ann Arbor, MI 48109, USA. E-mail address: [email protected] (C. Gerlein-Safdi). https://doi.org/10.1016/j.agrformet.2018.05.015 Received 2 August 2017; Received in revised form 10 May 2018; Accepted 21 May 2018 Available online 28 May 2018 0168-1923/ © 2018 Elsevier B.V. All rights reserved. C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

2018). There is growing interest in transpiration suppression during albedo and evaporative cooling effects can last for a considerable por- interactions between leaves and non-meteoric water (Barradas and tion of the day (Aparecido et al., 2017), shortening the duration of Glez-Medellín, 1999; Alvarado-Barrientos et al., 2014), but the im- water stress experienced by plants. portance of transpiration suppression to leaf water status and the spe- Dew deposition events are spatially extensive, and may occur si- cific manner by which dew deposition affects leaf energy, water, and multaneously over areas of thousands of squared kilometers (de Jeu carbon balances and the magnitude of these effects are still poorly re- et al., 2005). Even small changes in leaf energy, water, and carbon solved. balances on such scales can imply large landscape-scale impacts. The The leaf energy balance is often considered at steady state, a con- ubiquity of dew events across ecosystems and climates, including − dition when energy inputs due to net shortwave radiation, K (W m 2), during drought events, points to potentially important implications of −2 and incoming longwave radiation, Lin (W m ), are balanced by the transpiration suppression from dew for drought survival (Yang et al., −2 energy losses due to outgoing longwave radiation, Lout (W m ), sen- 2017) and plant community composition (Gauslaa, 2014; McLaughlin − sible heat transfers due to convection, H (W m 2), and latent energy et al., 2017). Indeed, transpiration suppression by dew can help plants transfers associated with the exothermic evaporation of liquid water maintain healthy leaf water status during soil droughts (Madeira et al., − within the leaf, λE (W m 2), where λ is the latent heat of vaporization 2002; Proctor, 2012). Key drivers of dew deposition, however, espe- − of water (J kg 1)(Gates, 1980; Vogel, 2012). The steady state energy cially air temperature and relative humidity, are expected to change balance can be expressed as (Campbell and Norman, 1998): rapidly with climate change (Cook et al., 2014), especially in tropical forests (Malhi and Wright, 2004; Nepstad et al., 2008), and will likely K +=++LλEHL (1) in out cause changes in the frequency and amount of dew deposition which can be simplified to: (Vuollekoski et al., 2015; Tomaszkiewicz et al., 2016). For example, reduced dew formation was a key feature of the 2005 mega-drought in R =+λE H, (2) n the Amazon (Frolking et al., 2011). Non-climate drivers of global −2 where Rn = K + Lin − Lout is the net radiation (W m ). Because the change such as deforestation also impact dew and fog formation by leaf energy loss processes are temperature-dependent, a steady state raising temperatures and lowering humidity (Moreira et al., 1997; Ray energy balance implies a steady state leaf temperature. In natural et al., 2006). The ecological impact of such changes in dew formation ecosystems, this leaf temperature is generally warmer than surrounding regimes is unclear. Although dew formation is included in many global air, due to the high radiative fluxes leaves experience, and their low climate models (Rosenzweig and Abramopoulos, 1997), its interaction thermal mass (Li et al., 2013). Smaller leaves tend to be at a tem- with vegetation carbon, energy, and water balance is not usually perature closer to the ambient air, because their boundary layer is evaluated, which represents a missing climate-ecosystem feedback in thinner, facilitating convective heat exchange (Givnish, 1987; Scoffoni evaluations of global change. However, including the interaction of et al., 2011). For this reason, and because high leaf temperatures can vegetation with non-meteoric water in global climate models requires impede photosynthesis or even result in leaf mortality, sun-exposed the development and testing of biophysical models of dew deposition, leaves are often smaller than shaded leaves (Gates, 1980). dew evaporation, and the consequences for leaf level carbon, energy The λE term in Eq. (1) couples steady-state leaf energy balance to and water balance. steady-state plant water balance, because evaporating water in leaves Some important components of such a model have been developed must be replaced by transport of water to the leaf via the xylem. Indeed, in different contexts: (i) informing the design of dew collection systems plants open their stomata to allow CO2 to diffuse in, with an inevitable to supplement water supply; and (ii) modeling effects of dew on agri- loss of water as water vapor diffuses out. This water loss leads to eva- cultural disease risks. Independently, these efforts have provided de- porative cooling, which plays an important role in balancing solar ra- tailed energy balance descriptions of synthetic condenser systems diative forcing (Pieruschka et al., 2010). When plant available water is (Nikolayev et al., 1996; Jacobs et al., 2002; Richards, 2009; Maestre- insufficient to meet lead water demand under ambient environmental Valero et al., 2012; Beysens, 2016) and the duration of leaf wetness on conditions, leaf transpiration rates are reduced through stomatal clo- plant leaves (Janssen and Römer, 1991; Evans et al., 1992; Sentelhas sure, which leads to a new steady-state energy balance with a higher et al., 2008; Schmitz and Grant, 2009; Bregaglio et al., 2010; Kim et al., leaf temperature. Thus, drought conditions, which restrict leaf water 2010). The aim of this study is to link insight from these models to each supply, elevate leaf temperature, causing a trade-off between water other in order to develop a representation of the leaf carbon, energy, conservation and temperature management (Rodriguez-Dominguez and water balance that provides a framework for interrogating the leaf- et al., 2016). Non-meteoric water on a leaf surface, however, increases - level physiological implications of dew deposition. To this end, we and externalizes – the water supply available for evaporative cooling propose a new leaf energy balance model that incorporates both the (Chu et al., 2012), potentially alleviating this tradeoff. In the presence deposition and the evaporation of dew. The energy balance model is of non-meteoric leaf water, the leaf energy balance equation is mod- coupled to a dynamic stomatal conductance and leaf carbon assimila- ified: tion model that tracks the effect of the presence of dew on leaf temperature, transpiration, and CO uptake. To test the model, we firstly R =++λE H E , (3) 2 ndew present the results of a calibration experiment that relates the voltage and evaporation of the dew (Edew) provides an additional leaf energy reading on commercially available capacitance leaf wetness sensors to sink. Applying this estimate to typical dew deposition volumes the mass of water accumulated on the surfaces of plant leaves exposed (0.5 mm/day (Monteith, 1963; Clus et al., 2008), with peak values of to the same conditions of humidity and temperature. The resulting 0.8 mm/day reported in some tropical forests (Andrade, 2003; Lakatos calibration relationship is then used to infer the mass of dew stored on a et al., 2012)) suggests that dew evaporation could satiate ≈1/6 of network of similar leaf wetness sensors deployed across a range of potential transpiration in tropical areas, less in arid climates (Monteith, elevation, vegetation, and topographic settings at Blue Oak Ranch Re- 1963). These estimates, however, neglect two additional impacts of serve, near San Jose California. Web camera images are used to identify dew deposition on leaves: (1) wet leaves have higher albedos, which a set of dew and low-density fog events, and leaf wetness sensor output reduces K (Pinter, 1986); and (2) wet leave have a lower temperature, is used to test model predictions based on locally observed temperature because of the cooling provided by dew evaporation which lowers and humidity during these events. The model is then applied to quantify

ﬂuxes of water (λE, Edew) and energy (H) from the leaf. Because dew transpiration suppression and changes in CO2 assimilation rates from formed before dawn can persist for up to 6 h after sunrise (Abtew and dew deposition. Melesse, 2012; Monteith, 1957) and dew may also form in the late afternoon before sunset (Wilson et al., 1999; Kabela et al., 2009),

306 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

2. Materials and methods emissivity of the surroundings as seen by the abaxial side of the leaf and

is here taken to be equal to ϵl. αdew and αdry are the albedo of a wet and 2.1. Modeling the effects of dew on a leaf a dry leaf (no units), respectively, σ is the Stefan-Boltzman constant in −2 −4 Wm K , and Tair is the temperature of the air in °C. For simplicity, 2.1.1. Energy balance the difference in albedo between wet and dry leaves is treated cate- Here, we present a process-based model of dew deposition on the gorically rather than a function of dew amount. A list of parameters is adaxial surface of the leaf and dew evaporation, and its effect on leaf available in Table 1. temperature, transpiration, and CO2 uptake by coupling the heat Using the Penman–Monteith equation, the evaporative cooling from equation to a dew mass balance equation (Richards, 2009; Vuollekoski transpiration is described as (Rodriguez-Iturbe and Porporato, 2004): et al., 2015). The heat equation describes the energy balance in terms of eTsat() leaf− eT c () air its effects on plant leaf temperature, as follows: λEρgλ= 0.622 v , air v P (6) d Tleaf ρ −3 ()Cmll+=+−− C dd m SR ln P lat SH l SλE l ,where air is the density of air in kg m , P is the atmospheric pressure dt (4) in Pa, esat is the saturation vapor pressure in Pa, ec is the vapor pressure λ −1 fi where Tleaf, Cl, and ml are the leaf temperature in °C, leaf specific heat in Pa, v is the latent heat of vaporization in J kg de ned as a −1 −1 λ 6 capacity in J kg K , and leaf mass in kg, respectively. Cd, and md are function of temperature (in °C) as (Dingman, 2002): v =10 × – −3 fi the specific heat capacity and mass of dew accumulated on the leaf. Sl is (2.5 2.36 × 10 × T), and gv is the total conductance de ned as: 2 λ −2 the surface area of the leaf in m , and Rn, H, and E are in W m , and gg g = hs , Plat is Edew (Eq. (3)) expressed in W. v gghs+ (7) The net radiation Rn is described as: with gs the stomatal conductance and gh the conductance of the −1 ⎧ (1−+αRdew ) sw (ϵ w ϵ atm ifmd > 0 boundary layer to water vapor, all three expressed in m s . The con- ⎪ 4 ductance of the boundary layer can be described as a function of the ⎪ ++−ϵlsurr ϵ )σT ( air 273.2) (ϵl ffi ⎪ 4 heat transfer coe cient, hc (Incropera et al., 2007; Schymanski et al., ⎪ ++ϵwleaf )σT ( 273.2) , Rn = 2013): ⎨(1−+αRdry ) sw ϵ l (ϵ atm ifmd = 0 ⎪ hc 4 g = ⎪ ++ϵsurr )σT ( air 273.2) h 2/3 ⎪ ρCNair aLe (8) ⎪ −+2 ϵσT ( 273.2)4 , ⎩ lleaf (5) −1 −1 with Ca is the heat capacity of air, equal to 1010 J kg K . NLe is the where the first term on the right hand side is the incoming shortwave dimensionless Lewis number, defined as the ratio of thermal diffusivity ff αa α ff radiation K, the second is the incoming longwave radiation Lin, and the to mass di usivity: NLe = , with a the thermal di usivity of air Dva 2 −1 third term is the outgoing longwave radiation Lout. ϵw (no units) is the (m s ) depending on the boundary layer temperature Tb = −7 −5 effective emissivity of dew water, ϵl (no units) is the effective emissivity (Tair + Tleaf)/2 as: αa =(Tb + 273.2) 1.32 × 10 − 1.73 × 10 and 2 −1 of the dry leaf, ϵatm (no units) is the effective emissivity of the atmo- Dva is the diffusivity of water vapor in air (m s ): Dva =(Tb + 273.2) −7 −5 sphere defined as (Kustas et al., 1527): 1.49 × 10 − 1.96 × 10 . The heat transfer coefficient hc 1/7 −1 −2 ϵatm =+0.642(eTc ( air )/( T air 273.2)) , and ϵsurr (no units) is the effective (W K m ) can be described as:

Table 1 List of parameters used in Section 2.

Parameter Value Unit Description Relevant equations

αdew 0.1 Albedo of a wet leaf (5)

αdry 0.5 Albedo of a dry leaf (5)

ϵw 0.97 Emissivity of water (5)

ϵl 0.95 Emissivity of a dry leaf (5) −6 −1 γ0 34.6 × 10 mol mol CO2 compensation point at T0 (15) −1 γ1 0.0451 K (15) −2 γ2 0.000347 K (15) −3 ρair 1.225 kg m Density of air (6), (8), (18) − − − σ 5.670373 × 10 8 Wm K 4 Stefan–Boltzman constant (5) −6 −1 cair 400 × 10 mol mol CO2 concentration of air (12), (20) −1 −1 Ca 1005 J kg K Heat capacity of air (8) −1 −1 Cl 3750 J kg K Leaf heat capacity (4) −1 −1 Cd 4181.3 J kg K Heat capacity of liquid water (4) −1 Hkc 59 430 J mol Activation energy for Kc (27) −1 Hko 36,000 J mol Activation energy for Ko (27) −1 Hvv 116,300 J mol Activation energy for Vc,max (26) −1 Hdv 202,900 J mol Deactivation energy for Vc,max (26) −1 Hvj 79,500 J mol Activation energy for Jmax (29) −1 Hdj 201,000 J mol Deactivation energy for Jmax (29) −2 Jmax,0 17.6 J m Electron transport capacity at T0 (29) −6 −1 Kc,0 302 × 10 mol mol Michaelis–Menten coeﬃcients for CO2 at T0 (27) −3 −1 Ko,0 256 × 10 mol mol Michaelis–Menten coeﬃcients for O2 at T0 (27) −1 MH2O 0.0182 kg mol Molar mass of water (19) NPr 0.71 Prandtl number for air (10) −1 Oc 0.21 mol mol Oxygen mole fraction in chloroplast (23) − − R 8.314 J mol 1 K 1 Ideal gas constant (13), (21), (25) −1 Sv 650 J mol Entropy term (26), (29)

T0 293.2 K Reference temperature (15), (26) −6 −2 −1 Vc,max,0 50 × 10 mol m s Maximum carboxylation rate at T0 (25), (26)

307 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

NNu is: hkca= , Ll (9) λT()dmdd ,if0dm > ⎧ vleaf dt dt where ka is the thermal conductivity of air in the boundary layer in Plat = −1 −1 ⎨λT()dmdd ,if0dm < WK m described as (Schymanski et al., 2013): ka =(T + 273.2) cleaf dt dt (17) − − ⎩ × 6.84 × 10 5 + 5.62 × 10 3, L is the characteristic length of the l λ leaf in m, taken to be equal to the leaf width (Schymanski and Or, with md the dew mass (kg) and c the latent heat of condensation −1 (J kg ) and is equal to λv. The dew mass balance can be written as 2016), and NNu is the dimensionless Nusselt number, the ratio of convective to conductive heat transfer across the boundary layer: follows:

N = 0.664 NN1/2 1/3, (10) d md ⎡ eTcair()− e satleaf ( T )⎤ Nu Re Pr = 0.622 Sρl air gh , dt ⎣ P ⎦ (18) with NPr the dimensionless Prandtl number, equal to 0.71 for air, and Note that in Eq. (18), the esat and the ec have opposite signs compared to NRe is the Reynolds number, defined as the ratio of inertial forces to viscous forces within a fluid: the expression of transpiration (Eq. (6)) because we are counting dew condensation positively and evaporation negatively. uLl NRe = , νair (11) 2.1.2. Modeling net assimilation −1 ffi fl with u the wind velocity in m s , and νair is the kinetic viscosity of the Water use e ciency (WUE) is the ratio of CO2 ux into the leaf over − air in m s 1, here taken to be a function of temperature using the the water vapor flux out of the leaf: ν empirical relation (Schymanski et al., 2013): air =(T + 273.2) × 9 × A − − n 10 8 − 1.13 × 10 5 WUE = , Emol (19) For the description of the stomatal conductance, we use an opti- fl −2 −1 mized expression of gs. Optimized stomatal conductance models aim at where Emol is the transpiration ux expressed in mol m s and reflecting the conflict between increased carbon uptake and limited which is easily calculated from λE as described in Eq. (6): λE water losses in plants (Wolf et al., 2016). In these models, stomatal Emol = , with MH2O the molar mass of water equal to λMvH2O −1 −2 −1 conductance is usually described as a function of carbon assimilation, 0.0182 kg mol . For An, the net flux of CO2 (mol m s ) we use the An, atmospheric CO2 concentration, cair, and water vapor deficit (VPD), model described in (Rodriguez-Iturbe and Porporato, 2004): D as a function of An (Vico et al., 2013; Medlyn et al., 2011). Al- cDair Agccn =−()air i , (20) though fairly new, these models have been overall very performant in sba describing stomatal response to environmental conditions during the where ci is the intercellular CO2 concentration within the leaf −1 −2 −1 day (Lloyd and Farquhar, 1994; Dewar et al., 2017). Here, we choose (mol mol ), and gsba (mol m s ) is the equivalent conductance (Vico et al., 2013) expressions for optimized stomatal aperture, gs,mol corresponding to the stomatal (gs), boundary layer (gh), and mesophyll −2 −1 (expressed in mol m s ) under rubisco and RuBP regeneration lim- (gm) conductances to CO2 put in series: itations: 1 g = sba 1.6RT 1.37 RT 1 An cair leaf,K leaf,K ⎧ , ifDA>> 0 andn 0, under rubisco limitation ++ cD1.6 λ gPs gPhm g (21) ⎪ air m gs,mol = An 1.6 Γ* ⎨ 3×>> , ifDA 0 andn 0, under RuBP regeneration limitation with atmospheric pressure P in Pa, T the leaf temperature in K, and ⎪ cDair λm leaf,K ⎪ ⎩0, otherwise gh and gs the boundary layer and stomatal conductances to water vapor −1 (12) expressed in m s and as described in Eqs. (8) and (13), respectively. − − The mesophyll conductance to CO can be described as a function of Stomatal conductance in mol m 2 s 1 can then be transformed to 2 − leaf temperature using the following empirical relation (Buckley et al., ms 1 as: 2014): RT g = leaf,K g 2 s s,mol (13) ⎡ Tleaf ⎤ P gm = 0.18 exp 0.71027 ln⎛ ⎞ ⎢ ⎝ 36.75 ⎠ ⎥ (22) −2 −1 ⎣ ⎦ In this expression, An is expressed in mol m s and described later in o −2 −1 Eq. (31), cair is the CO2 concentration of air, set here to with Tleaf in C and gm in mol m s . In addition, An can also be −6 −1 400 × 10 mol mol , and λm is the marginal water use efficiency described as the minimum of the rubisco-limited and the light-limited cair −6 −1 taken to be: λmm,0= λ with cair,0 = 380 × 10 mol mol and assimilation rates. The rubisco limitation can be described as cair,0 −6 −1 (Rodriguez-Iturbe and Porporato, 2004): λm,0 = 400 × 10 mol mol (Vico et al., 2013). D (no unit) is defined as: ci − Γ* AVn,c= c,max − Rl, eT()− eT () cKic++(1 OK co / ) (23) D = max⎛ 0, sat leaf c air ⎞ ⎝ P ⎠ (14) and the RuBP regeneration limitation caused by light limitation: Γ* with P, esat, and ec as described above. Finally, is the CO2 compen- ci − Γ* −1 AJn,q = − Rl, sation point (mol mol ) calculated as a function of leaf temperature 4(ci + 2Γ*) (24) T in K as: leaf,K * −1 Γ is the CO2 compensation point (mol mol ) calculated as a function Γ* =+γγTTγTT[1 ( −+− ) ( )2 ]. −2 −1 01leaf,K 0 2 leaf,K 0 (15) of leaf temperature as described in Eq. (15). Rl (mol m s ) is the non-photorespiratory CO release in the light and is calculated as a The sensible heat flux in and out of the leaf is described as: 2 function of leaf temperature following the expression (Schymanski and HhTT=−2(cleafair ), (16) Or, 2016): fl which is positive (energy ux out of the leaf) when the leaf is hotter 46.39 R = 0.0089V exp⎡ 18.72 − ⎤ than the ambient air, and negative (flux into the leaf) when the leaf is l c,max,0 ⎢ −3 ⎥ ⎣ 10 RTleaf,K ⎦ (25) cooler than the air. Finally, when dew is present, the latent energy −2 −1 dissipated from evaporating dew or accumulated from condensing dew Vc,max is the maximum carboxylation rate (mol m s ) described as:

308 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

HT surface properties of leaves such as hydrophobicity and total heat ca- exp⎡ vv 1 − 0 ⎤ ⎣ RT0 ()Tleaf,K ⎦ pacity. They have a dielectric constant that is altered by the presence of Vc,max= V c,max,0 , ST− H 1exp+ ⎡ v leaf,K dv ⎤ water on the sensor surface (Letts and Mulligan, 2005; Gil et al., 2011; ⎣ RTleaf,K ⎦ (26) Berkelhammer et al., 2013). The dielectric constant can be calibrated −1 against the mass of water on the sensor surface for a given excitation Kc and Ko are the Michaelis–Menten coeﬃcients (mol mol ) for CO2 voltage, providing quantitative information about water inputs and and O2, respectively and are described as (where x is either c or o): leaf-level storage (Cobos, 2013). What remains unclear is the robustness

⎡ Hkx ⎛ T0 ⎞⎤ of the relationships between the sensor output, the evolution of the K = K exp ⎜⎟1 − . xx,0⎢ ⎥ water mass on the sensor surface, and the water mass present on plant ⎣ RT0 ⎝ Tleaf,K ⎠⎦ (27) leaves under the same conditions. Thus, before interpreting the leaf J is the electron transport rate given by the lower root of the quadratic wetness sensor network outputs in terms of masses of water accumu- equation: lating on the sensors – and thus potentially surrounding plant leaves – a 2 calibration experiment is necessary. κ1JκQJJκQJ−+()2 max + 2 max = 0, (28) withκκ12== 0.95, 0.25 and, 2.2.1. Field data An array of over 57 wireless sensor nodes is deployed at the Blue Hvj T0 exp⎡ 1 − ⎤ Oak Ranch Reserve (BORR), a Biological Field Station and Ecological ⎣ RT0 ()Tleaf,K ⎦ Jmax= J max,0 , Reserve located near San Jose, California (Hamilton et al., 2011). Leaf STv leaf,K − Hdj 1exp+ ⎡ ⎤ wetness data, air temperature and relative humidity are reported to- ⎣ RTleaf,K ⎦ (29) gether at 25 of the 57 nodes, with 15 minutes resolution (Fig. S1 in the −2 −1 and Q (mol photons m s ) the absorbed photon irradiance, which Supporting Information). Data is collected and transmitted wirelessly can be calculated from incoming shortwave radiation as (Buckley et al., using an eKo Pro wireless network (eK2120, MEMSIC, Inc, Andover, 2014; Schymanski and Or, 2016): Q =×10−6 (1− αR ) sw , with α taken 0.5666 MA). Leaf wetness sensors were installed progressively from late 2012 α α to be either dew or dry depending on whether dew is present or not. through 2014. Incoming solar radiation, wind speed, rainfall, and at- To calculate An, we replace ci in Eqs. (23) and (24) by its expression mospheric pressure are available hourly at six weather stations begin- derived from Eq. (20). Isolating An, we obtain two expressions for An,c ning operation in 2010. Because BORR experiences fog events in ad- and An,q (Schymanski and Or, 2016) dition to dew, 2011–2013 web camera pictures taken from the Lick

Oc gsba Rl Vc,max Observatory, situated on the summit of Mount Hamilton, were used to AcKn,c= ⎛ air++ c 1 ⎞ −+ ⎝ ()Ko ⎠ 2 2 2 classify time periods when fog was visible over the reserve. Images were 2 1 Oc Oc taken every 15 min. Eight categories of imagery were used in the ⎜⎟⎛⎛ ⎞ ⎞ ⎛⎜⎛ ⎞ − cKair++ c 1212Γ gsba + Rl − cKair−+− c * gsba 2 ⎝⎝ ()Ko ⎠ ⎠ ⎝ ⎝ ()Ko ⎠ classification, including dark, fog, clear day, or lens obscured. Foggy ⎞ days were separated into conditions suitable for radiation or advection +−2RVlc,maxc,max⎟ V ⎠ fog based on wind speed and leaf wetness patterns. Radiation fog forms − gsba Rl J 1 Acn,q=+(2Γ air *) −+ under low windspeed conditions (< 2.23 m s ), and is controlled by 2 28 1 2 similar physics to that of to dew formation. Leaf wetness sensor output −++−−+−16((cgRcgRJJair 2Γ*))(8(4Γsba l air *)8)sba l 8 during the classified radiation fog events were similar to those observed (30) for dew events. They were characterized by a single wetting event, with and a a slow water deposition phase and a rapid drying phase (Fig. 1). Advection fog is usually accompanied by higher windspeed and gen- AAAnn= min(,cn ,,q ) (31) erates multiple small drying and rewetting events. To test the dew model, we focused on five leaf wetness sensors that had data over- 2.2. Model testing lapping the fog classification data and low data gap percentage. We included both dew events and conditions when radiative fog occurred To test the dew deposition and evaporation components of the de- in the model testing. The data is available online through the UC Ber- rived model, we considered a suite of data collected on a leaf wetness keley Sensor Database website (http://sensor.berkeley.edu/aboutDB. sensor network deployed at the Blue Oak Ranch Reserve near San Jose, html). California. Leaf wetness sensors were originally developed to identify risks of plant foliar disease under wet leaf conditions (Chungu et al., 2.2.2. Leaf wetness sensor calibration 2001; Dalla Marta et al., 2005, 2007; Rossi et al., 2008; Gil et al., 2011). Calibration experiments were conducted within a a Contemporary leaf wetness sensors mimic key thermodynamic and 30 × 50 × 50 cm plexiglass chamber with a removable front panel. At

Fig. 1. Leaf wetness sensor (black solid line) and windspeed (red dotted line) data for three diﬀerent water deposition events: advection fog on the left, radiation fog in the center, and dew on the right. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.)

309 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316 the bottom of the panel, an ultrasonic mist generator (NHKR, M-011) repeat measures, and propagated into all subsequent computations with > 300 mL/h capacity was floated in a water reservoir, and covered using the R qpcR package. Where the total mass of water accumulated with a wire screen to prevent water splashing from the reservoir into on the leaf surface during the experimental trial fell within the 95% the chamber. Mist generated at the bottom of the chamber was dis- confidence interval for the leaf dry mass (i.e. a low signal to noise ratio, persed using a waterproof fan (JSarci, Scenario 120). A dielectric leaf primarily occurring on leaves with a small surface area), the data was wetness sensor (Decagon, 40020) was mounted horizontally 30 cm excluded from subsequent analysis. above the floor of the chamber (c. 27 cm above the reservoir water surface), run using an excitation voltage of 3000 mV, and reported 2.3. Model experiments continuously to a datalogger (Campbell CR1000). Leaves of Quercus douglasii (blue oak) and Quercus lobata (valley We first look at the influence of leaf size and dew deposition amount oak) were placed horizontally one at a time within the wire cradle in a on transpiration and CO2 assimilation. Meteorological data for a typical dry chamber. These two species were chosen because they are com- day at BORR were chosen (Fig. S2 in the Supporting Information) and monly found at the BORR research station. The wire cradle was sus- the model was run for different characteristic leaf lengths, Ll (Eq. (9)). pended at the same level as the leaf wetness sensor. The cradle was We vary the dew deposition amount at dawn and model transpiration suspended from a load cell (Measurement Specialties, Inc., FN3280) and assimilation during a full diurnal cycle. We compare these results to located outside the chamber. Mass readings from the load cell were the transpiration and assimilation of a dry leaf in the same environ- continuously logged throughout the experiment, and the load cell was mental conditions. calibrated at the beginning and end of each experimental run. Once the We then explore the effects of a changing climate by varying re- load cell readings stabilized, the ultrasonic mist generator and fan were lative humidity, wind speed and air temperature. We use the diurnal started, the chamber was sealed, and the load cell and wetness sensor time series of air temperature, pressure, and solar radiation presented readings logged over 45 min. Visually, the chamber reached a homo- above (Fig. S2 in the Supporting Information) and run the model for a ‘ ’ − geneously foggy condition within 45 s of starting the experiment. set of wind speeds from 0.1 to 5 m s 1 and relative humidities from 1 to While the experiment ran, we measured the dry weight of a polyvinyl 100%. Relative humidity and windspeed are kept constant for the entire alcohol cloth. At the conclusion of the experiment, we swabbed both diurnal cycle. We set the dew deposition amount at dawn to 0.4 kg/m2 surfaces of the leaf wetness sensor with the cloth and weighed the cloth and calculate the decrease in transpiration and CO2 assimilation asso- to estimate the mass of water accumulated on the sensor. ciated with the presence of dew, compared to diurnal transpiration and We measured the surface area of the tested leaves as follows: the assimilation levels of a dry leaf. We also record leaf wetness duration. fi fl leaves were rstly attened and fastened to a white background with a To reproduce the warming expected in the Tropics by the end of the 2 one cm reference square marked on it in black. The sheets were 21st century (Cook et al., 2014), we repeat the experiment with an air scanned in an optical scanner, converted to a binary (black and white temperature 5 °C higher than recorded air temperature. The experiment image) and the black pixel count associated with the leaf used to esti- is conducted for leaf characteristic lengths Ll (Eq. (9)) of 1 and 50 cm. mate the surface area. Seven and eight different leaves were tested for for Q. douglasii and Q. lobata, respectively. Control runs were also made 3. Results and discussion with no leaf in place to allow the mass change associated with fog deposition on the cradle to be determined. The mass changes in the 3.1. Laboratory leaf wetness measurements absence of leaves were negligible, suggesting that no significant deposition of water on the wire cradle occurred. Fig. 2 shows a scatter plot of the mass of water deposited on the leaf The logged data were subjected to a series of quality assurance wetness sensor at the end of the experimental runs and the voltage steps. The mass calibration performed at the beginning and end of each output from the sensor recorded at the same time period. The strong trial was used to correct for sensor drift during the trial by imposing a linear trend between the water mass on the sensor and the output, had a linear drift term that was subtracted from the measured data. 95% coefficient of determination of 0.98 and was well described by the confidence intervals on the load cell measurements were made based on following relationship:

M =×±×(1.11 10−−35 34.20 10 ) LWS −±×(0.313 1.8 10−2 ) (32)

where M is the total mass of water deposited on the leaf wetness sensor (g), and LWS represents the output from the leaf wetness sensor in mV. Error terms indicate the 95% confidence intervals that would apply to predictions made with this fit, and indicate that the mass of water accumulating on the leaf wetness sensor could be estimated from the output voltage with an error of approximately 5%. The experimental results agree with previous analyses comparing sensor water accumulation to sensor output (Cobos, 2013) and indicate that the dielectric sensor provided robust detection of leaf mass changes on the sensor surface. Unlike previous results (Cobos, 2013), all data used in devel- oping this relationship were derived from (i) independent experimental runs, and (ii) from the ambient deposition of mist droplets on the sensor surface, rather than a sprayed application of water to the sensor surface. We postulate that these conditions may be more representative of field conditions induced by fog or dew formation than direct spraying. Water deposition on the real leaves shows a strong linear relation- Fig. 2. Calibration curve for the mass of deposited water on the leaf wetness ship with water on the leaf wetness sensor (Fig. 3) under the same sensor as a function of the voltage reported by the leaf wetness sensor. The conditions. Overall, we find that the leaf wetness sensor and the leaves 2 relationship was strongly linear and well described (R = 0.97) by the function: accumulate c. the same amount of water, but the leaf wetness sensor −3 − Mass(g) = 1.105 × 10 Voltage(mV) 0.3126. appears to systematically underestimate the amount of water

310 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

Fig. 3. Water accumulated on the leaf wetness sensor and on real leaves of Quercus douglasii and Quercus lobata. The different shades of grey show the different trials done using different leaves of each species (n = 7 for Q. douglasii and n = 8 for Q. lobata). accumulated on real leaves. These differences are likely linked to dif- therefore lead to lower-than-expected water deposition amount on the ferences in leaf heat capacity and surface properties that are improperly sensor which are not captured by the model. reproduced by the sensor. 3.3. Model predictions 3.2. Testing the dew model against field data 3.3.1. Leaf size effects on transpiration and assimilation The dew deposition model was tested using field data for 16 dif- Transpiration suppression in leaves lasts until dew evaporates, ferent dew and radiative fog events at all 5 sensors. Field data for air which depends on evaporation rate and on the amount of dew accu- temperature, relative humidity, incoming solar radiation, wind speed, mulated before sunrise. We find that transpiration suppression over the and atmospheric pressure were used as input in the model. The data course of a day (Fig. S2) is near-linearly correlated with dew amount for was linearly interpolated to 1s time resolution. The model solved for all leaves sizes (Fig. 5). Unexpectedly, we find that carbon assimilation leaf temperature and dew amount, and dew accumulation was com- is also decreased by the presence of dew (Fig. 6), and that for small pared to estimated accumulation from the leaf wetness sensors. Eq. (32) leaves, the decrease follows a similarly linear relationship to what was was adapted to the field sensors, which all showed a baseline (mV found for transpiration. In addition, both transpiration and assimilation output for dry sensor) of 0.4884, instead of 0.313 as described in Sec- are further decreased in larger leaves, likely because larger leaves stay tion 3.1. This is due to the difference in datalogger used and does not wet longer, thanks to a deeper boundary layer that effectively reduces affect the measurement (Chris Chambers, Meter Group, Inc., personal dew evaporation. The effect of dew deposition on the assimilation in communication). Model performance was estimated using an absolute larger leaves appears to level off for dew amounts larger than c. 0.6 kg/ 2 root-mean-square error (RMSE), and relative RMSE (normalized by the m . This non-linearity is linked to midday stomatal closure in larger maximum recorded leaf wetness). Results are shown in Table 2. leaves (Fig. S2), likely due to high leaf temperature and low relative The model was generally in very good agreement with the recorded humidity in the middle of the day (Tenhunen et al., 1981). Finally, we leaf wetness, capturing the timing of dew deposition, and the amount find that WUE is not affected by the presence of dew (data not shown). deposited (Fig. 4). While only a limited number of dew event were available, the model represented these events better that radiative fog 3.3.2. Effects of climatic changes events (Table 2, average RMSE/normalized RMSE for dew events: In Fig. 7, we show how wind speed, relative humidity, and leaf

0.018/0.42; fog events: 0.045/0.79), likely because some subtleties of characteristic length impact transpiration, CO2 assimilation, and dew radiative fog formation, such as aerosol concentration (Yamaguchi duration. As expected, for small leaves, high relative humidity is asso- et al., 2013) or fog water content (Katata et al., 2011), are omitted from ciated with an increase in transpiration and CO2 assimilation suppres- the model. The model errs towards overestimating the amount of water sion, which corresponds with longer leaf wetness duration. In the limit deposited on the sensor. This is consistent with the results observed in of RH = 100% and when Tair and Tleaf are equal, as it is often the case in our laboratory experiment, where leaf wetness sensor deposition was the early morning, dew is not excepted to evaporate at all, with leaf systematically lower than water accumulating on real leaves (Fig. 3). In wetness lasting multiple hours. Similarly, lower wind speed leads to addition to issues with the inherent design of the leaf wetness sensor, higher decrease in transpiration because of the associated increase in there might also be additional explanations for this: (1) the model aims boundary layer thickness (Eq. (8)), which increases the diffusion time of at representing a real, transpiring leaf, whereas the leaf wetness sensor water vapor out of the boundary layer. CO2 assimilation is also de- is an unanimated object. Leaf transpiration will decrease leaf tem- creased at low windspeeds for small leaves. perature further than what would be expected for the leaf wetness In large leaves (Fig. 7, bottom row) we actually observe an increase sensor. However, for all the dew and radiation fog events observed in in transpiration and CO2 assimilation at low windspeed. This is again the data, the water deposition started at night, when stomata are shut, likely linked to an increase in stomatal conductance due to the presence limiting the effect of this potential source of error. (2) Sensor place- of dew at mid-day, when larger leaves would usually close their stomata ment: in the field, the leaf wetness sensors are attached to the radiation (Figs. 5, 6, and S2). At high relative humidity and windspeed below − shield (see Fig. S1 in the Supporting Information). The sensors are 1ms 1,wefind an increase in assimilation that also corresponds to a therefore in thermal equilibrium with the radiation shield, which is decrease in transpiration. This will therefore lead to an overall increase likely to be generally warmer than what the leaf wetness sensor would in water use efficiency (Eq. (19)). This domain of optimal dew effect is have been if it had been installed in suspension, like a real leaf. This will associated with an increase in CO2 assimilation of up to c. 20% and a

311 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

Fig. 4. Modeling dew formation for two dew events at the Blue Oak Ranch Reserve in California. The red line shows the modeled dew amount, obtained from air temperature, wind speed, relative humidity, solar radiation, and atmospheric pressure data. The black line shows the leaf wetness sensor output at the site, with the mV output transformed to dew amount using the relation from Section 3.1. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) decrease in transpiration by about 15%. Interestingly, this effect does where high relative humidity leads to common dew deposition (Lakatos not appear in smaller leaves (Fig. 7, top row). Fig. S3 (in Supporting et al., 2012; Aparecido et al., 2017). Such levels could also have a Information) explores the effect that an increase of air temperature by significant impact on plant transpiration in the case of drought which is 5 °C would have on transpiration, assimilation and dew duration. a big unknown in our understanding of vegetation response to a Overall, the patterns observed are similar to the ones in Fig. 7, but a changing climate (Fisher et al., 2017). However, leaves under low water clear decrease in the effects of dew is seen, with a general decrease in potential are likely to close their stomata to prevent cavitation transpiration and CO2 assimilation reduction by about 5 percentage (Rodriguez-Dominguez et al., 2016), which would then eliminate points. transpiration along with any potential effect from dew through transpiration suppression. The drought level at which stomata closure will 3.4. Implications happen is highly dependent on the species (Brodribb and Holbrook, 2003) and the model proposed here can easily be modified to explore ff ff ff Dew deposition is a phenomenon that is commonplace across most the e ects of dew on di erent species and in di erent types of eco- ecosystems on the planet (Vuollekoski et al., 2015). A large knowledge systems. gap exists in vegetation response to a changing climate (Fisher et al., The negative impact of dew on carbon assimilation, which experi- 2017; Kooperman et al., 2018), and as dew frequency and amount are ences up to 10% decrease under high dewfall, is mostly due to stomata ff expected to be affected by climate change (Vuollekoski et al., 2015; closure under (1) decreased incoming radiation which a ects assim- Tomaszkiewicz et al., 2016), understanding the effects of these changes ilation by decreasing the absorbed photon irradiance, Q and therefore J for plants growing in areas of common dewfall may become even more through Eq. (28), and (2) low leaf temperatures that are extended after important. Our results point at significant water savings from tran- sunrise when the leaf is covered in dew. In the case of amphistomatous ff spiration suppression of up to a 25% decrease from daily transpiration leaves, dew might also directly a ect CO2 concentration in the chlor- oplasts by obstructing part of the stomata and hindering CO2 diffusion. levels, in particular for bigger leaves with Ll around 50 cm (Fig. 5). At low wind speed, bigger leaves might even experience a combined de- Here, we model the case of a hypostomatous leaf with dew depositing crease in transpiration and increase in assimilation leading to an in- on the adaxial surface, and this third mechanism can therefore be ne- crease in WUE (Fig. 7). Leaves this size are common in tropical areas, glected.

Fig. 5. Left: total daily change in transpiration due to the presence of dew as a function of dew amount present on the leaf at sunrise for four diﬀerent leaf characteristic lengths, Ll. Right: Relative change in daily transpiration due to the presence of dew compared to transpiration in a dry leaf. The environmental parameters used to drive the model are presented in the Supporting Information (Fig. S2).

312 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

Fig. 6. Left: total daily change in CO2 assimilation due to the presence of dew as a function of dew amount present on the leaf at sunrise for four diﬀerent leaf characteristic lengths, Ll. Right: Relative change in daily CO2 assimilation due to the presence of dew compared to assimilation in a dry leaf. The environmental parameters used to drive the model are presented in the Supporting Information (Fig. S2).

One could have expected that transpiration suppression from dew rainfall interception happens primarily in the upper leaves of the plants, allowed leaves to open up their stomata to uptake CO2 with little water which are also the leaves most exposed to the sun and more likely to losses. Our model indicates that this is only the case for large leaves in transpire during the day. This might lead to a transpiration suppression low windspeed and high relative humidity conditions, and that this effect larger than what we estimated here for dew. However, dew for- positive effect on assimilation does not happen in smaller leaves, al- mation is larger during nights with clear skies. On the opposite, rain though experimental evidence is still needed to confirm this result. clouds will reduce incoming radiation to the leaf, and as a consequence However, it is important to note that WUE appears to be mostly un- reduce the effects of leaf wetness on transpiration and assimilation changed by the presence of dew, showing that transpiration and as- compared to that of dew. How these two aspects of rainfall interception similation are decreased by a similar factor due to the presence of dew. would balance out is still unclear, and should be the topic of a more Finally, the results presented here are also applicable to other detailed investigation. phenomena, such as rainfall interception. Thick canopies can intercept One of the main limitations of the present model is that it represents the integrity of small rainfall events (Guevara-Escobar et al., 2007; dew deposition as a uniform film of water on top of the leaf. Indeed, Zimmermann et al., 2013), leading to leaf water deposition at the dew might also deposit on the abaxial surface of the leaf. However, the surface of the leaves much larger than in the case of dew. In addition, emissivity of the surrounding environment seen by the bottom of the

Fig. 7. Expected change in daily transpiration and CO2 assimilation (% change from daily values in the absence of dew), as well as dew dry-out time (in min since − midnight) for varying wind speeds (m s 1) and relative humidity values (%), using the air temperature and solar radiation data presented in Fig. S2 assuming a dew 2 deposition amount at dawn of 0.4 kg/m . Top row: Ll = 1 cm, bottom row: Ll = 50 cm. Fig. S3 shows the same plots generated for an air temperature 5 °C higher than the temperature presented in Fig. S2.

313 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

Table 2 Table showing the results of the model tested on 16 diﬀerent dew and fog events at the Blue Oak Ranch Reserve in California. The normalized RMSE shows the RMSE divided by the maximum recorded by the leaf wetness sensor for each event. All times shown are local (Paciﬁc Standard) time.

Date Type RMSE Normalized RMSE Maximum leaf wetness (kg/m2) Time of max. wetness Time wetting starts Time wetting ends

Sensor Model Sensor Model Sensor Model Sensor Model

2012/08/31 Fog 0.020 0.43 0.046 0.073 8:30 9:24 4:30 4:21 11:07 11:28 2012/10/13 Fog 0.110 2.35 0.049 0.240 8:30 9:17 23:21 21:00 10:59 11:27 2013/04/02 Fog 0.070 0.72 0.099 0.230 6:00 8:13 19:00 20:24 11:59 10:59 2013/04/25 Fog 0.060 0.79 0.077 0.20 5:16 7:31 20:18 20:41 11:31 10:03 2013/05/17 Dew 0.007 0.14 0.051 0.053 6:43 7:15 0:00 1:22 8:24 8:20 2013/05/22 Dew 0.007 0.31 0.022 0.009 4:33 2:56 23:45 0:58 6:23 4:52 2013/05/23 Dew 0.050 1.05 0.050 0.110 1:40 7:37 21:38 21:51 8:23 9:37 2013/05/27 Fog 0.008 0.39 0.020 0.034 6:43 7:16 3:20 2:06 8:38 8:13 2013/06/19 Fog 0.014 0.25 0.056 0.063 6:57 7:29 2:36 3:37 8:38 8:42 2013/06/26 Fog 0.062 1.02 0.061 0.170 6:45 7:20 0:00 0:00 9:45 9:22 2013/08/09 Dew 0.008 0.17 0.046 0.042 6:57 7:42 3:56 3:59 8:38 8:44 2013/08/26 Fog 0.031 0.58 0.054 0.110 6:57 8:18 2:25 2:29 10:12 10:07 2013/09/05 Fog 0.012 0.32 0.039 0.021 7:26 8:18 1:58 1:57 9:17 9:28 2013/09/30 Fog 0.030 0.35 0.084 0.088 6:30 8:50 19:00 2:40 11:15 10:27 2013/10/30 Fog 0.096 1.87 0.052 0.225 6:00 9:13 20:20 20:17 10:45 11:42 2013/12/19 Fog 0.026 0.41 0.063 0.093 3:45 8:18 21:33 21:30 9:57 10:00

leaf is larger than the emissivity of the clear sky seen by the top of the experiencing an enhanced impact due to the increased leaf wetness leaf (see Table 1). Dew deposition is therefore much more frequent on duration. In addition, CO2 assimilation is negatively impacted by the the adaxial surface of the leaf than the abaxial one. In addition, most presence of dew, because of the decrease in leaf temperature associated species have no stomata on their adaxial surface (Aparecido et al., with it. However, we found that at high humidity and low wind speeds, 2017). These two elements combined lead us to the conclusion that large dew-wetted leaves can experience both a decrease in transpiration modeling dew deposition on the top of the leaf only is valid in most and an increase in assimilation, leading to an increase water use effi- cases. In the case of amphistomatous leaves, dew deposition would clog ciency. Here we represent dew deposition as a film of water, and while up the stomata of the adaxial surface. However, the diffusion of CO2 in we recognize that this model has its limitations, we expect that a more water is 104 times slower than in air, and wet stomata can therefore be detailed representation of dew droplets as half-spheres would further considered to be impermeable to CO2. Other minor effects, such as free increase transpiration suppression. This work is only the first step to- convection for large leaves have been neglected here because it hasn’t wards understanding the impact of water deposition on leaves. Future been found to be a significant heat transfer mechanism in most leaves work should strive to tease out the relative importance of the effects of (Leuning, 1988) and would likely not impact the results significantly. temperature, radiation, and stomata clogging on both transpiration and In addition, a more accurate model would represent dew droplets as assimilation. In particular, understanding the mechanistic pathways of half-spheres (Beysens, 1995; Tanaka, 2002; Beysens et al., 2006), with these interactions is critical to incorporate the effects of dew deposition dynamic nucleation processes throughout dewfall and dew re-eva- into ecosystem-scale models. poration (Beysens, 2005; Leach et al., 2006). These processes vary greatly between species and are complex to model (Goldsmith et al., Acknowledgments 2016), but a more precise description of the dew would capture subtleties that might have a large overall effect on transpiration suppres- The authors thank Michael Hamilton for all his help, as well as the sion. For example, we expect that spherical drops would increase sur- University of California, Blue Oak Ranch Reserve, UC Natural Reserve face roughness, further deepening the leaf boundary layer, and further System, which was supported by an award from the National Science decreasing transpiration (Schuepp, 1993). Similarly, we do not account Foundation – Grant 0934296. CGS and KKC acknowledge the financial for localized increases in relative humidity due to the re-evaporation of support of NASA Headquarters under the NASA Earth and Space dew on the leaf, or from surrounding leaves. This would again have the Science Fellowship Program – Grant 14-EARTH14F-241 – and of a Mary net effect of decreasing leaf transpiration. Finally, leaf angle is expected and Randall Hack ’69 Graduate Award and the Science, Technology, to play a major role in both dew amount and incoming radiation. In this and Environmental Policy Fellowship from the Princeton model, we assumed a flat leaf, because dew beading up and rolling off Environmental Institute. CGS thanks Missy Holbrook and the depart- the leaf will be highly dependent on the surface of the leaf: pubescence ment of Organismic and Evolutionary Biology at Harvard University for will increase dewfall and retain dew droplets (Konrad et al., 2014), hosting her during part of this work. whereas hydrophobic leaves will have low water retention (Holder, 2012a,b). These subtleties will be exacerbated in non-horizontal leaves Appendix A. Supplementary data and should be taken into account when evaluating the effects of dew deposition for a specific species, but are beyond the scope of this work. Supplementary data associated with this article can be found, in the Leaves are also known to adopt a more vertical orientation at the online version, at https://doi.org/10.1016/j.agrformet.2018.05.015. hottest hours of the day to decrease incoming radiation from the sun (Ehleringer and Comstock, 1987; King, 1997; Posada et al., 2012). This References phenomenon generally occurs in the afternoon (Sellers, 2003), at a time when dew has already evaporated, and we would therefore expect that Abtew, W., Melesse, A., 2012. Evaporation and Evapotranspiration. Springer Science & the relative decrease in morning transpiration due to dew would be Business Media, Dordrecht. Agam, N., Berliner, P.R., 2006. Dew formation and water vapor adsorption in semi-arid even larger if leaf angle dynamics were taken into account. environments – a review. J. Arid Environ. 65 (4), 572–590. In conclusion, our results show that dew deposition can have a large Alvarado-Barrientos, M.S., Holwerda, F., Asbjornsen, H., Dawson, T.E., Bruijnzeel, L.A., impact on transpiration suppression. In particular, we find that tran- 2014. Suppression of transpiration due to cloud immersion in a seasonally dry Mexican weeping pine plantation. Agric. Forest Meteorol. 186, 12–25. spiration suppression effects vary with leaf size, with larger leaves

314 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

Andrade, J.L., 2003. Dew deposition on epiphytic bromeliad leaves: an important event in Gotsch, S.G., Nadkarni, N., Darby, A., Glunk, A., Mackenzie, D., Davidson, K., Dawson, a Mexican tropical dry deciduous forest. J. Trop. Ecol. 19 (5), 479–488. T.E., 2015. Life in the treetops: ecophysiological strategies of canopy epiphytes in a Aparecido, L.M.T., Miller, G.R., Cahill, A.T., Moore, G.W., 2017. Leaf surface traits and tropical montane cloud forest. Ecol. Monogr. 85 (3), 393–412. water storage retention affect photosynthetic responses to leaf surface wetness among Guevara-Escobar, A., González-Sosa, E., Véliz-Chávez, C., Ventura-Ramos, E., Ramos- wet tropical forest and semiarid savanna plants. Tree Physiol. 37 (10), 1285–1300. Salinas, M., 2007. Rainfall interception and distribution patterns of gross precipita- Barradas, V.L., Glez-Medellín, M.G., 1999. Dew and its effect on two heliophile under- tion around an isolated Ficus benjamina tree in an urban area. J. Hydrol. 333 (2–4), storey species of a tropical dry deciduous forest in Mexico. Int. J. Biometeorol. 43 532–541. (1), 1–7. Hamilton, M.P., Dawson, T.E., Thompson, S.E., 2011. The very large ecological array. Berkelhammer, M., Hu, J., Bailey, A., Noone, D.C., Still, C.J., Barnard, H., Gochis, D., AGU Fall Meeting Abstracts, Abstract B14A-08. Hsiao, G.S., Rahn, T., Turnipseed, A., 2013. The nocturnal water cycle in an open- Hill, A.J., Dawson, T.E., Shelef, O., Rachmilevitch, S., 2015. The role of dew in Negev canopy forest. J. Geophys. Res.: Atmosp. 118 (1), 10225. Desert plants. Oecologia 178 (2), 317–327. Beysens, D., 1995. The formation of dew. Atmosp. Res. 39 (1), 215–237. Holder, C.D., 2012a. Effects of leaf hydrophobicity and water droplet retention on canopy Beysens, D., 2005. Dew nucleation and growth. Comptes Rendus Phys. 7 (9), 1082–1100. storage capacity. Ecohydrology 6 (3), 483–490. Beysens, D., 2016. Estimating dew yield worldwide from a few meteo data. Atmosp. Res. Holder, C.D., 2012b. The relationship between leaf hydrophobicity, water droplet re- 167 (C), 146–155. tention, and leaf angle of common species in a semi-arid region of the western United Beysens, D., Steyer, A., Guenoun, P., Fritter, D., Knobler, C.M., 2006. How does dew States. Agric. Forest Meteorol. 152, 11–16. form? Phase Trans. 31 (1–4), 219–246. Holloway, P.J., 1994. chap. Plant cuticles: physiochemical characteristics and biosynth- Bregaglio, S., Donatelli, M., Confalonieri, R., Acutis, M., Orlandini, S., 2010. Multi metric esis. In: Percy, K., Cape, C.N., Jagels, R., Simpson, C.J. (Eds.), Air Pollutants and the evaluation of leaf wetness models for large-area application of plant disease models. Leaf Cuticle. Springer, Heidelberg, Germany. Agric. Forest Meteorol. 151 (9), 1163–1172. Incropera, F.P., DeWitt, D.P., Bergman, T.L., Lavine, A.S., 2007. Fundamentals of Heat Brodribb, T., Holbrook, N., 2003. Stomatal closure during leaf dehydration, correlation and Mass Transfer, 6th ed. John Wiley, Hoboken, NJ. with other leaf physiological traits. Plant Physiol. 132 (4), 2166–2173. Jacobs, A.F.G., Heusinkveld, B.G., Berkowicz, S.M., 2002. A simple model for potential Buckley, T.N., Martorell, S., Diaz-Espejo, A., Tomas, M., Medrano, H., 2014. Is stomatal dewfall in an arid region. Atmosp. Res. 64 (1), 285–295. conductance optimized over both time and space in plant crowns? A field test in Jacobs, A.F.G., Heusinkveld, B.G., Wichink Kruit, R.J., Berkowicz, S.M., 2006. grapevine (Vitis vinifera). Plant Cell Environ. 37 (12), 2707–2721. Contribution of dew to the water budget of a grassland area in the Netherlands. Water Campbell, G.S., Norman, J.M., 1998. An Introduction to Environmental Biophysics. Resour. Res. 42 (3). Springer New York, New York, NY. Janssen, L.H.J.M., Römer, F.G., 1991. The frequency and duration of dew occurrence over Chu, H.-S., Chang, S.-C., Klemm, O., Lai, C.-W., Lin, Y.-Z., Wu, C.-C., Lin, J.Y., Jiang, J.Y., a year: model results compared with measurements. Tellus B 43 (5), 408. Chen, J., Gottgens, J.F., Hsia, Y.-J., 2012. Does canopy wetness matter? Jordan, D.N., Smith, W.K., 1994. Energy balance analysis of nighttime leaf temperatures Evapotranspiration from a subtropical montane cloud forest in Taiwan. Hydrol. and frost formation in a subalpine environment. Agric. Forest Meteorol. 71 (3), Process. 28 (3), 1190–1214. 359–372. Chungu, C., Gilbert, J., Townley-Smith, F., 2001. Septoria tritici blotch development as Kabela, E.D., Hornbuckle, B.K., Cosh, M.H., Anderson, M.C., Gleason, M.L., 2009. Dew affected by temperature, duration of leaf wetness, inoculum concentration, and host. frequency, duration, amount, and distribution in corn and soybean during SMEX05. Plant Dis. 85 (4), 430–435. Agric. Forest Meteorol. 149 (1), 11–24. Clus, O., Ortega, P., Muselli, M., Milimouk, I., Beysens, D., 2008. Study of dew water Katata, G., Kajino, M., Hiraki, T., Aikawa, M., 2011. A method for simple and accurate collection in humid tropical islands. J. Hydrol. 361 (1), 159–171. estimation of fog deposition in a mountain forest using a meteorological model. J. Cobos, D.R., 2013. Predicting the Amount of Water on the Surface of the LWS Dielectric Geophys. Res. 116 (20), D20102. Leaf Wetness Sensor. Decagon Devices, Inc. Application Note 509-332-5600. Kim, K.S., Taylor, S.E., Gleason, M.L., Nutter Jr., F.W., Coop, L.B., Pfender, W.F., Seem, Cook, B.I., Smerdon, J.E., Seager, R., Coats, S., 2014. Global warming and 21st century R.C., Sentelhas, P.C., Gillespie, T.J., Marta, A.D., Orlandini, S., Dalla Marta, A., 2010. drying. Climate Dyn. 43 (9–10), 2607–2627. Spatial portability of numerical models of leaf wetness duration based on empirical Dalla Marta, A., Magarey, R.D., Martinelli, L., Orlandini, S., 2007. Leaf wetness duration approaches. Agric. Forest Meteorol. 150 (7–8), 871–880. in sunfl( ower Helianthus annuus): analysis of observations, measurements and simu- King, D.A., 1997. The functional significance of leaf angle in Eucalyptus. Aust. J. Bot. 45 lations. Eur. J. Agron. 26 (3), 310–316. (4), 619–639. Dalla Marta, A., Magarey, R.D., Orlandini, S., 2005. Modelling leaf wetness duration and Konrad, W., Burkhardt, J., Ebner, M., Roth-Nebelsick, A., 2014. Leaf pubescence as a downy mildew simulation on grapevine in Italy. Agric. Forest Meteorol. 132 (1–2), possibility to increase water use efficiency by promoting condensation. Ecohydrology 84–95. 8 (3), 480–492. de Jeu, R.A.M., Holmes, T.R.H., Owe, M., 2005. Determination of the effect of dew on Kooperman, G.J., Chen, Y., Hoffman, F.M., Koven, C.D., Lindsay, K., Pritchard, M.S.,

passive microwave observations from space. Rem. Sens. Agric. 5976, 51–59. Swann, A.L.S., Randerson, J.T., 2018. Forest response to rising CO2 drives zonally Dewar, R., Mauranen, A., M“akel”a, A., H“oltt”a, T., Medlyn, asymmetric rainfall change over tropical land. Nat. Climate Change 8 (April (5)), B., Vesala, T., 2017. New insights into the covariation of stomatal, mesophyll and 434–440. hydraulic conductances from optimization models incorporating nonstomatal lim- Kustas, W.P., Rango, A., Uijlenhoet, R., 1994. A simple energy budget algorithm for the itations to photosynthesis. New Phytol. 217 (October (2)), 571–585. snowmelt runoff model. Water Resour. Res. 30 (May (5)), 1515–1527. Dingman, S.L., 2002. Physical Hydrology, 2nd ed. Waveland Press, Inc. Lakatos, M., Obregón, A., B”udel, B., Bendix, J., 2012. Midday dew – an over- Ehleringer, J.R., Comstock, J., 1987. Leaf absorptance and leaf angle: mechanisms for looked factor enhancing photosynthetic activity of corticolous epiphytes in a wet stress avoidance. In: Tenhunen, J.D., Catarino, F.M., Lange, O.L., Oechel, W.C. (Eds.), tropical rain forest. New Phytol. 194 (1), 245–253. Plant Response to Stress: Functional Analysis in Mediterranean Ecosystems. Springer Leach, R.N., Stevens, F., Langford, S.C., Dickinson, J.T., 2006. Dropwise condensation: Berlin Heidelberg, Berlin, Heidelberg, pp. 55–76. http://dx.doi.org/10.1007/978-3- experiments and simulations of nucleation and growth of water drops in a cooling 642-70868-8_3. system. Langmuir 22 (21), 8864–8872. Evans, K.J., Nyquist, W.E., Latin, R.X., 1992. A model based on temperature and leaf Letts, M.G., Mulligan, M., 2005. The impact of light quality and leaf wetness on photo- wetness duration for establishment of alternaria leaf-blight of muskmelon. synthesis in north-west Andean tropical montane cloud forest. J. Trop. Ecol. 21 (05), Phytopathology 82 (8), 890–895. 549–557. Fisher, J.B., Melton, F., Middleton, E., Hain, C., Anderson, M., Allen, R., McCabe, M.F., Leuning, R., 1988. Leaf temperatures during radiation frost Part II. A steady state theory. Hook, S., Baldocchi, D., Townsend, P.A., Kilic, A., Tu, K., Miralles, D.D., Perret, J., Agric. Forest Meteorol. 42 (2), 135–155. Lagouarde, J.-P., Waliser, D., Purdy, A.J., French, A., Schimel, D., Famiglietti, J.S., Li, S., Zhang, Y.-J., Sack, L., Scoffoni, C., Ishida, A., Chen, Y.-J., Cao, K.-F., 2013. The Stephens, G., Wood, E.F., 2017. The future of evapotranspiration: global require- heterogeneity and spatial patterning of structure and physiology across the leaf ments for ecosystem functioning, carbon and climate feedbacks, agricultural man- surface in giant leaves of Alocasia macrorrhiza. PLoS ONE 8 (6), e66016. 13 agement, and water resources. Water Resour. Res. 53 (4), 2618–2626. Lloyd, J., Farquhar, G.D., 1994. C discrimination during CO2 assimilation by the ter- Frolking, S., Milliman, T., Palace, M., Wisser, D., Lammers, R., Fahnestock, M., 2011. restrial biosphere. Oecologia 99 (3–4), 201–215. Tropical forest backscatter anomaly evident in SeaWinds scatterometer morning Madeira, A.C., Kim, K.S., Taylor, S.E., Gleason, M.L., 2002. A simple cloud-based energy overpass data during 2005 drought in Amazonia. Rem. Sens. Environ. 115 (3), balance model to estimate dew. Agric. Forest Meteorol. 111 (1), 55–63. 897–907. Maestre-Valero, J.F., Ragab, R., Martínez-Alvarez, V., Baille, A., 2012. Estimation of dew Gao, L., McCarthy, T.J., 2006. Contact angle hysteresis explained. Langmuir 22 (14), yield from radiative condensers by means of an energy balance model. J. Hydrol. 6234–6237. 460–461 (C), 103–109. Gates, D., 1980. Biophysical Ecology. Springer-Verlag New York. Malhi, Y., Wright, J., 2004. Spatial patterns and recent trends in the climate of tropical Gauslaa, Y., 2014. Rain, dew, and humid air as drivers of morphology, function and rainforest regions. Philos. Trans. R. Soc. B: Biol. Sci. 359 (1443), 311–329. spatial distribution in epiphytic lichens. Lichenologist 46 (01), 1–16. McLaughlin, B.C., Ackerly, D.D., Klos, P.Z., Natali, J., Dawson, T.E., Thompson, S.E., Gerlein-Safdi, C., Gauthier, P.P.G., Caylor, K.K., 2018. Dew-induced transpiration sup- 2017. Hydrologic refugia, plants, and climate change. Global Change Biol. 66, pression impacts the water and isotope balances of Colocasia leaves. Oecologia. 107–121. Gil, R., Bojaca, C.R., Schrevens, E., 2011. Suitability evaluation of four methods to esti- Medlyn, B.E., Duursma, R.A., Eamus, D., Ellsworth, D.S., Prentice, I.C., Barton, C.V.M., mate leaf wetness duration in a greenhouse rose crop. Acta Horticult. 797–804. Crous, K.Y., De Angelis, P., Freeman, M., Wingate, L., 2011. Reconciling the optimal Givnish, T.J., 1987. Comparative studies of leaf form: assessing the relative roles of se- and empirical approaches to modelling stomatal conductance. Global Change Biol. 17 lective pressures and phylogenetic constraints. New Phytol. 106 (S1), 131–160. (6), 2134–2144. Goldsmith, G.R., Bentley, L.P., Shenkin, A., Salinas, N., Blonder, B., Martin, R.E., Castro- Monteith, J.L., 1957. Dew. Q. J. R. Meteorol. Soc. 83, 322–341. Ccossco, R., Chambi-Porroa, P., Diaz, S., Enquist, B.J., Asner, G.P., Malhi, Y., 2016. Monteith, J.L., 1963. In: Rutter, A.J., Whitehead, F.H. (Eds.), The Water Relations of Variation in leaf wettability traits along a tropical montane elevation gradient. New Plants. Blackwell Scientific, New York, pp. 37–56. Phytol. 214 (3), 989–1001. Moreira, M., Sternberg, L., Martinelli, L., Victoria, R., Barbosa, E., Bonates, L., Nepstad,

315 C. Gerlein-Safdi et al. Agricultural and Forest Meteorology 259 (2018) 305–316

D., 1997. Contribution of transpiration to forest ambient vapour based on isotopic Sellers, B., 2003. Diurnal ﬂuctuations and leaf angle reduce glufosinate eﬃcacy. Weed measurements. Global Change Biol. 3 (5), 439–450. Technol. 17 (2), 302–306. Neinhuis, C., Barthlott, W., 1997. Characterization and distribution of water-repellent, Sentelhas, P.C., Dalla Marta, A., Orlandini, S., Santos, E.A., Gillespie, T.J., Gleason, M.L., self-cleaning plant surfaces. Ann. Bot. 79 (6), 667–677. 2008. Suitability of relative humidity as an estimator of leaf wetness duration. Agric. Nepstad, D.C., Stickler, C.M., Soares-Filho, B., Merry, F., 2008. Interactions among Forest Meteorol. 148 (3), 392–400.

Amazon land use, forests and climate: prospects for a near-term forest tipping point. Tanaka, K., 2002. Multi-layer model of CO2 exchange in a plant community coupled with Philos. Trans. R. Soc. B: Biol. Sci. 363 (1498), 1737–1746. the water budget of leaf surfaces. Ecol. Modell. 147 (1), 85–104. Nikolayev, V.S., Beysens, D., Gioda, A., Milimouka, I., Katiushin, E., Morel, J.P., 1996. Tenhunen, J.D., Lange, O.L., Braun, M., 1981. Midday stomatal closure in Mediterranean Water recovery from dew. J. Hydrol. 182 (1), 19–35. type sclerophylls under simulated habitat conditions in an environmental chamber – Pan, Y., Wang, X., 2014. Effects of shrub species and microhabitats on dew formation in a II. Effect of the complex of leaf temperature and air humidity on gas exchange of revegetationstabilized desert ecosystem in Shapotou, northern China. J. Arid Land 6 Arbutus unedo and Quercus ilex. Oecologia 50 (August (1)), 5–11. (4), 389–399. Tolk, J.A., Howell, T.A., Steiner, J.L., Krieg, D.R., Schneider, A.D., 1995. Role of tran- Pieruschka, R., Huber, G., Berry, J.A., 2010. Control of transpiration by radiation. Proc. spiration suppression by evaporation of intercepted water in improving irrigation Natl. Acad. Sci. U. S. A. 107 (30), 13372–13377. efficiency. Irrig. Sci. 16 (2), 89–95. Pinter Jr., P.J., 1986. Effect of dew on canopy reflectance and temperature. Rem. Sens. Tomaszkiewicz, M., Najm, M.A., Beysens, D., Alameddine, I., Zeid, E.B., El-Fadel, M., Environ. 19 (2). 2016. Projected climate change impacts upon dew yield in the Mediterranean basin. Posada, J.M., Siev“anen, R., Messier, C., Perttunen, J., Nikinmaa, E., Lechowicz, Sci. Total Environ. 566–567, 1339–1348. M.J., 2012. Contributions of leaf photosynthetic capacity, leaf angle and self-shading Uclés, O., Villagarcía, L., Moro, M.J., Canton, Y., Domingo, F., 2013. Role of dewfall in the to the maximization of net photosynthesis in Acer saccharum: a modelling assess- water balance of a semiarid coastal steppe ecosystem. Hydrol. Process. 28 (4), ment. Ann. Bot. 110 (3), 731–741. 2271–2280. Proctor, M.C.F., 2012. Dew, where and when? ’There are more things in heaven and Vico, G., Manzoni, S., Palmroth, S., Weih, M., Katul, G., 2013, December. A perspective on

earth, Horatio, than are dreamt of in your philosophy’. New Phytol. 194 (1), 10–11. optimal leaf stomatal conductance under CO2 and light co-limitations. Agric. Forest Ray, D.K., Nair, U.S., Lawton, R.O., Welch, R.M., Pielke, R.A., 2006. Impact of land use on Meteorol. 182–183, 191–199. Costa Rican tropical montane cloud forests: sensitivity of orographic cloud formation Vogel, S., 2012. The Life of a Leaf. The University of Chicago Press. to deforestation in the plains. J. Geophys. Res.: Atmosp. 111 (D), D02108. Vuollekoski, H., Vogt, M., Sinclair, V.A., Duplissy, J., J”arvinen, H., Kyrö, E.-M., Richards, K., 2009. Adaptation of a leaf wetness model to estimate dewfall amount on a Makkonen, R., Petäjä, T., Prisle1, N.L., Räisänen, P., Sipilä, M., Ylhäisi, J., Kulmala, roof surface. Agric. Forest Meteorol. 149 (8), 1377–1383. M., 2015. Estimates of global dew collection potential on artificial surfaces. Hydrol. Rodriguez-Dominguez, C.M., Buckley, T.N., Egea, G., de Cires, A., Hernandez-Santana, V., Earth Syst. Sci. 19 (1), 601–613. Martorell, S., Diaz-Espejo, A., 2016. Most stomatal closure in woody species under Wilson, T.B., Bland, W.L., Norman, J.M., 1999. Measurement and simulation of dew ac- moderate drought can be explained by stomatal responses to leaf turgor. Plant Cell cumulation and drying in a potato canopy. Agric. Forest Meteorol. 93 (2), 111–119. Environ. 39 (9), 2014–2026. Wolf, A., Anderegg, W.R.L., Pacala, S.W., 2016. Optimal stomatal behavior with com- Rodriguez-Iturbe, I., Porporato, A., 2004. Ecohydrology of Water-Controlled Ecosystems. petition for water and risk of hydraulic impairment. PNAS 113 (46), E7222–E7230. Cambridge University Press, New York. Xu, Y., Yan, B., Tang, J., 2015. The effect of climate change on variations in dew amount Rosenzweig, C., Abramopoulos, F., 1997. Land-surface model development for the GISS in a paddy ecosystem of the Sanjiang Plain, China. Adv. Meteorol. 2015 (2), 1–9. GCM. J. Climate 10 (8), 2040–2054. Yamaguchi, T., Noguchi, I., Watanabe, Y., Katata, G., Sato, H., Hara, H., 2013. Aerosol Rossi, V., Caffi, T., Giosu, S., Bugiani, R., 2008. A mechanistic model simulating primary deposition and behavior on leaves in cool-temperate deciduous forests. Part 2: infections of downy mildew in grapevine. Ecol. Modell. 212 (3–4), 480–491. characteristics of fog water chemistry and fog deposition in Northern Japan. Asian J. Schmitz, H.F., Grant, R.H., 2009. Precipitation and dew in a soybean canopy: spatial Atmosp. Environ. 7 (1), 8–16. variations in leaf wetness and implications for Phakopsora pachyrhizi infection. Agric. Yang, X.-D., Lv, G.-H., Ali, A., Ran, Q.-Y., Gong, X.-W., Wang, F., Liu, Z.-D., Qin, L., Liu, Forest Meteorol. 149 (10), 1621–1627. W.-G., 2017. Experimental variations in functional and demographic traits of Lappula Schuepp, P.H., 1993. Leaf Boundary Layers: Tansley Review no. 59 (Vol. 125). New semiglabra among dew amount treatments in an arid region. Ecohydrology 10 (6), Phytologist. e1858. Schymanski, S.J., Or, D., 2016. Wind increases leaf water use efficiency. Plant Cell Yates, D.J., Hutley, L.B., 1995. Foliar uptake of water by wet leaves of Sloanea woollsii,an Environ. 39 (7), 1448–1459. Australian subtropical rainforest tree. Aust. J. Bot. 43 (2), 157–167. Schymanski, S.J., Or, D., Zwieniecki, M., 2013. Stomatal control and leaf thermal and Zangvil, A., 1996. Six years of dew observations in the Negev Desert, Israel. J. Arid hydraulic capacitances under rapid environmental fluctuations. PLoS ONE 8 (1), Environ. 32 (4), 361–371. e54231. Zimmermann, B., Zimmermann, A., Scheckenbach, H.L., Schmid, T., Hall, J.S., van Scoffoni, C., Rawls, M., McKown, A., Cochard, H., Sack, L., 2011. Decline of leaf hydraulic Breugel, M., 2013. Changes in rainfall interception along a secondary forest succes- conductance with dehydration: relationship to leaf size and venation architecture. sion gradient in lowland Panama. Hydrol. Earth Syst. Sci. 17 (11), 4659–4670. Plant Physiol. 156 (2), 832–843.

316